Morgan Stanley Fund Management has observed that the central focus of the artificial intelligence sector over the last two years has revolved around the question of model intelligence. However, as 2026 approaches, a notable shift is underway: enhancements in model capability are beginning to cross the threshold for practical commercial usability, allowing AI to evolve from a simple chat and content creation tool into a digital employee capable of independently handling complex jobs.

The newest generation of frontier models is demonstrating marked improvements in long-horizon tasks, software engineering, and scientific research. Whereas previous models excelled at standardized jobs requiring only minutes to complete, the latest versions can now operate continuously for hours, autonomously invoking tools, executing assignments, checking results, and refining their approach. This transition is expanding AI's value proposition from augmenting human effort to actually completing the work itself. High-value scenarios that previously resisted large-scale automation, such as software development, cybersecurity, drug design, chip engineering, and scientific exploration, are now being unlocked.

Current industry materials indicate that the sector's core focus has shifted from chasing raw token usage to evaluating whether models can fulfill complex engineering and genuine business objectives. Concurrently, competition among foundational models is experiencing a different kind of change, characterized by a narrowing capability gap and steadily declining prices. Domestic models are rapidly advancing in areas like coding and agent functionality, while open-source offerings are increasingly handling mature workloads. Over the long term, establishing a permanent competitive moat by relying solely on a single top-tier model is becoming more difficult. What enterprise clients will consistently pay for may no longer be the model itself, but rather the comprehensive working systems created when models are integrated with corporate data, permission structures, business processes, and specialized knowledge.

Industry insiders cited in the materials note that as models become more commoditized, the value of proprietary data, contextual information, workflow designs, and organizational frameworks actually increases. This carries significant implications for the investment thesis in the computer sector. While the market has historically concentrated on large model companies and computing infrastructure, the application layer is now poised to enter a phase of revenue realization. Software firms possessing deep industry data, access to core business systems, and control over customer workflows are better positioned to convert AI from a feature enhancement into a sustainable new revenue model. This is especially true in categories like office software, enterprise management, cybersecurity, and software development, where agents have the potential to redesign entire business processes rather than simply adding a chat interface.

The media and content sector also warrants attention. Falling inference costs and improving model capabilities will continue to reduce production expenses for images, videos, gaming assets, and advertising materials. Yet companies that ultimately thrive must own valuable intellectual property, user access points, distribution channels, and monetization opportunities. AI is more likely to amplify the productivity and supply capacity of established premium content platforms than to eliminate the content industry altogether. In light of these trends, the firm believes AI investment is now entering its second act. The first phase rewarded investments in stronger models and greater computational power, while the second phase requires identifying which companies can translate cheaper and more capable intelligence into tangible revenue and profits. Looking ahead, agent commercialization, the iterative release of domestic models, and the growing AI revenue contributions from software and content providers are likely to be the most compelling areas to monitor within the computer and media sectors.